Merge pull request django-haystack#790 from andrewschoen/feature/haystack-identifier-module

Added a new setting, HAYSTACK_IDENTIFIER_METHOD, which will allow a cust...

Author: andrewschoen

AUTHORS

@@ -70,3 +70,4 @@ Thanks to
     * Erik Rose (erikrose) for a quick pyelasticsearch-compatibility patch
     * Stefan Wehrmeyer (stefanw) for a simple search filter fix
     * Dan Watson (dcwatson) for various patches.
+    * Andrew Schoen (andrewschoen) for the addition of ``HAYSTACK_IDENTIFIER_METHOD``

docs/settings.rst

@@ -268,3 +268,20 @@ An example::
     HAYSTACK_DJANGO_ID_FIELD = 'my_django_id'
 
 Default is ``django_id``.
+
+
+``HAYSTACK_IDENTIFIER_METHOD``
+==============================
+
+**Optional**
+
+This setting allows you to provide a custom method for
+``haystack.utils.get_identifier``. Useful when the default identifier
+pattern of <app.label>.<object_name>.<pk> isn't suited to your
+needs.
+
+An example::
+
+    HAYSTACK_IDENTIFIER_METHOD = 'my_app.module.get_identifier'
+
+Default is ``haystack.utils.default_get_identifier``.

haystack/utils/__init__.py

@@ -1,33 +1,77 @@
 import re
+
 from haystack.constants import ID, DJANGO_CT, DJANGO_ID
 from haystack.utils.highlighting import Highlighter
 
 
-IDENTIFIER_REGEX = re.compile('^[\w\d_]+\.[\w\d_]+\.\d+$')
+from django.conf import settings
 
+try:
+    from django.utils import importlib
+except ImportError:
+    import importlib
 
-def get_model_ct(model):
-    return "%s.%s" % (model._meta.app_label, model._meta.module_name)
+IDENTIFIER_REGEX = re.compile('^[\w\d_]+\.[\w\d_]+\.\d+$')
 
 
-def get_identifier(obj_or_string):
+def default_get_identifier(obj_or_string):
     """
     Get an unique identifier for the object or a string representing the
     object.
-    
+
     If not overridden, uses <app_label>.<object_name>.<pk>.
     """
     if isinstance(obj_or_string, basestring):
         if not IDENTIFIER_REGEX.match(obj_or_string):
-            raise AttributeError("Provided string '%s' is not a valid identifier." % obj_or_string)
-        
+            raise AttributeError(u"Provided string '%s' is not a valid identifier." % obj_or_string)
+
         return obj_or_string
-    
-    return u"%s.%s.%s" % (obj_or_string._meta.app_label, obj_or_string._meta.module_name, obj_or_string._get_pk_val())
+
+    return u"%s.%s.%s" % (
+        obj_or_string._meta.app_label,
+        obj_or_string._meta.module_name,
+        obj_or_string._get_pk_val()
+    )
+
+
+def _lookup_identifier_method():
+    """
+    If the user has set HAYSTACK_IDENTIFIER_METHOD, import it and return the method uncalled.
+    If HAYSTACK_IDENTIFIER_METHOD is not defined, return haystack.utils.default_get_identifier.
+
+    This always runs at module import time.  We keep the code in a function
+    so that it can be called from unit tests, in order to simulate the re-loading
+    of this module.
+    """
+    if not hasattr(settings, 'HAYSTACK_IDENTIFIER_METHOD'):
+        return default_get_identifier
+
+    module_path, method_name = settings.HAYSTACK_IDENTIFIER_METHOD.rsplit(".", 1)
+
+    try:
+        module = importlib.import_module(module_path)
+    except ImportError:
+        raise ImportError(u"Unable to import module '%s' provided for HAYSTACK_IDENTIFIER_METHOD." % module_path)
+
+    identifier_method = getattr(module, method_name, None)
+
+    if not identifier_method:
+        raise AttributeError(
+            u"Provided method '%s' for HAYSTACK_IDENTIFIER_METHOD does not exist in '%s'." % (method_name, module_path)
+        )
+
+    return identifier_method
+
+
+get_identifier = _lookup_identifier_method()
+
+
+def get_model_ct(model):
+    return "%s.%s" % (model._meta.app_label, model._meta.module_name)
 
 
 def get_facet_field_name(fieldname):
     if fieldname in [ID, DJANGO_ID, DJANGO_CT]:
         return fieldname
-    
+
     return "%s_exact" % fieldname

tests/core/custom_identifier.py

@@ -0,0 +1,7 @@
+
+def get_identifier_method(key):
+    """
+    Custom get_identifier method used for testing the
+    setting HAYSTACK_IDENTIFIER_MODULE
+    """
+    return key

tests/core/tests/utils.py

@@ -1,6 +1,7 @@
 from django.test import TestCase
+from django.test.utils import override_settings
 from haystack.utils import log
-from haystack.utils import get_identifier, get_facet_field_name, Highlighter
+from haystack.utils import get_identifier, get_facet_field_name, Highlighter, _lookup_identifier_method
 from core.models import MockModel
 
 
@@ -16,45 +17,58 @@ def test_get_facet_field_name(self):
 class GetFacetFieldNameTestCase(TestCase):
     def test_get_identifier(self):
         self.assertEqual(get_identifier('core.mockmodel.1'), 'core.mockmodel.1')
-        
+
         # Valid object.
         mock = MockModel.objects.get(pk=1)
         self.assertEqual(get_identifier(mock), 'core.mockmodel.1')
 
+    @override_settings(HAYSTACK_IDENTIFIER_METHOD='core.custom_identifier.get_identifier_method')
+    def test_haystack_identifier_method(self):
+        get_identifier = _lookup_identifier_method()
+        self.assertEqual(get_identifier('a.b.c'), 'a.b.c')
+
+    @override_settings(HAYSTACK_IDENTIFIER_METHOD='core.custom_identifier.not_there')
+    def test_haystack_identifier_method_bad_path(self):
+        self.assertRaises(AttributeError, _lookup_identifier_method)
+
+    @override_settings(HAYSTACK_IDENTIFIER_METHOD='core.not_there.not_there')
+    def test_haystack_identifier_method_bad_module(self):
+        self.assertRaises(ImportError, _lookup_identifier_method)
+
 
 class HighlighterTestCase(TestCase):
     def setUp(self):
         super(HighlighterTestCase, self).setUp()
         self.document_1 = "This is a test of the highlightable words detection. This is only a test. Were this an actual emergency, your text would have exploded in mid-air."
         self.document_2 = "The content of words in no particular order causes nothing to occur."
         self.document_3 = "%s %s" % (self.document_1, self.document_2)
-    
+
     def test_find_highlightable_words(self):
         highlighter = Highlighter('this test')
         highlighter.text_block = self.document_1
         self.assertEqual(highlighter.find_highlightable_words(), {'this': [0, 53, 79], 'test': [10, 68]})
-        
+
         # We don't stem for now.
         highlighter = Highlighter('highlight tests')
         highlighter.text_block = self.document_1
         self.assertEqual(highlighter.find_highlightable_words(), {'highlight': [22], 'tests': []})
-        
+
         # Ignore negated bits.
         highlighter = Highlighter('highlight -test')
         highlighter.text_block = self.document_1
         self.assertEqual(highlighter.find_highlightable_words(), {'highlight': [22]})
-    
+
     def test_find_window(self):
         # The query doesn't matter for this method, so ignore it.
         highlighter = Highlighter('')
         highlighter.text_block = self.document_1
-        
+
         # No query.
         self.assertEqual(highlighter.find_window({}), (0, 200))
-        
+
         # Nothing found.
         self.assertEqual(highlighter.find_window({'highlight': [], 'tests': []}), (0, 200))
-        
+
         # Simple cases.
         self.assertEqual(highlighter.find_window({'highlight': [0], 'tests': [100]}), (0, 200))
         self.assertEqual(highlighter.find_window({'highlight': [99], 'tests': [199]}), (99, 299))